Skip to content

ag2

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

ag2

README.md

AG2 Mypy tool Agent with A2A Protocol

This sample demonstrates an AG2 agent that is exposed through the A2A protocol. It shows how an AG2 can communicate using A2A as a lingua franca, which is helpful for building a distributed agent system where different agents can communicate with each other using the A2A protocol. The agent-as-a-service architecture allows you to encapsulate agent logic, the local environment, tools, and other capabilities into a separate service that can be reused in different agent workflows.

Here is a sequence diagram that shows how the current demo implementation of the A2A protocol works:

sequenceDiagram
    participant Client as A2A Client
    participant Server as A2A Server
    participant Agent as AG2 (LLM + MCP Client)
    participant Tool as Mypy Tool

    Client->>Server: Send task with query
    Server->>Agent: Forward query to AG2 agent
    Note over Server,Agent: Real-time status updates (streaming)

    Agent->>Agent: LLM decides to use tool
    Agent->>Tool: Send tool execution request
    Tool->>Agent: Return tool results

    Agent->>Agent: LLM processes tool results
    Agent->>Server: Return completed response
    Server->>Client: Respond with task results
Loading

Possible use cases for Agent-as-a-Service architecture

  • Tool Access: Leverage various MCP or custom tools for complex tasks.
  • Web Browsing: Access to web browsing capabilities.
  • Code Execution: Run Python code for data analysis tasks.
  • Image Generation: Create images from text descriptions.
  • Real-time Streaming: Get status updates during processing.
  • Cross-Framework Communication: Demonstrates A2A's ability to connect different agent frameworks.

The demo

Here we have a simple demo that shows how to use the A2A protocol to communicate with an AG2 agent. We have

  • one A2A-served remote agent a2a_python_reviewer.py
  • two different A2A clients, which communicate with the remote agent using the A2A protocol: CLI code generator cli_codegen_a2a_client.py and FastAPI code generator fastapi_codegen_a2a_client.py

Prerequisites

  • Python 3.12 or higher
  • UV package manager
  • OpenAI API Key (for default configuration)

Setup & Running

  1. Navigate to the samples directory:

    cd samples/python/agents/ag2

  2. Create an environment file with your API key (uses openai gpt-4o):

    echo "OPENAI_API_KEY=your_api_key_here" > .env

  3. Install the dependencies:

    uv sync

  4. Run the remote agent:

    uv run a2a_python_reviewer.py

  5. In a new terminal, start an A2AClient interface to interact with the remote (ag2) agent. You can use one of the following clients:

    • Method A: Run the CLI client

      uv run cli_codegen_a2a_client.py

    • Method B: Run the FastAPI client

      uv run fastapi_codegen_a2a_client.py

Learn More

Disclaimer

Warning

The sample code provided is for demonstration purposes only. When building production applications, it is critical to treat any agent operating outside of your direct control as a potentially untrusted entity.

All data received from an external agent—including but not limited to its AgentCard, messages, artifacts, and task statuses—should be handled as untrusted input. For example, a malicious agent could provide an AgentCard containing crafted data in its fields (e.g., description, name, skills.description). If this data is used without sanitization to construct prompts for a Large Language Model (LLM), it could expose your application to prompt injection attacks. Failure to properly validate and sanitize this data before use can introduce security vulnerabilities into your application.

Developers are responsible for implementing appropriate security measures, such as input validation and secure handling of credentials, to protect their systems and users.